It has been demonstrated that circRNAs are abundant and stable in exosomes, suggesting their significant translational potential as circulating biomarkers for cancer diagnosis
[8]. In hepatocellular carcinoma, exosomal
circPTGR1 was shown to promote cancer progression through the regulation of the miR-449a/MET pathway
[9]. Similarly,
circNRIP1 was also proven to be transmitted via exosomes and promoted tumorigenesis and metastasis of gastric cancer
[10]. In laryngeal squamous cell carcinoma,
circRASSF2 was secreted by exosomes and promoted tumor growth through the regulation of the miR-302b-3p/IGF-1R pathway
[11]. High
circCNOT2 expression was associated with poor progression-free survival of patients with breast cancer, and
circCNOT is detectable in cell-free RNAs from patient plasma samples
[12]. In addition, circRNAs can also be detected in circulating tumor cells
[13]. Furthermore, circRNAs have been shown to be highly enriched in blood platelets compared with nucleated cells, which can be used for cancer diagnosis
[14]. Since protein carriers such as high-density lipoprotein and Argonaute 2 transport miRNAs
[15], the circulation of circRNAs might also be mediated by certain protein carriers or RNA-binding proteins.
3. Functional Roles of Circulating circRNAs in Lung Cancer
3.1. Cancer Progression
The presence of
F-circEA generated from the
EML–ALK fusion gene was verified in non-small cell lung cancer (NSCLC) cells and in the plasma of NSCLC patients
[16].
F-circEA promoted cancer cell migration and invasion, suggesting that
F-circEA could be a novel liquid biopsy marker for NSCLC. Through circRNA profiling of serum or plasma obtained from patients, the clinical relevance of many circulating circRNAs has been explored.
Global circRNA expression can be profiled by using RNA sequencing (RNA-seq) followed by bioinformatic approaches
[17].
circFARSA was identified as an upregulated circRNA in NSCLC tissues compared with adjacent normal tissue by analyzing back-spliced reads on RNA-seq data
[18].
circFARSA expression was higher in the plasma from NSCLC patients than in that from healthy volunteers and showed a good diagnostic value for NSCLC (AUC = 0.71). cDNA encoding
circFARSA was cloned into the pLCDH-ciR vector, which was specifically designed to overexpress circular transcripts
[19].
circFARSA overexpression enhanced the migration and invasion of A549 cells. Through in silico analyses, circFARSA was predicted to sponge miR-330 and miR-326 and regulate fatty acid synthesis. This is one of the earliest studies investigating the possibility of plasma circRNAs as new biomarkers for NSCLC patients; however, it lacks functional evidence supporting the molecular mechanism of
circFARSA in NSCLC.
Through a microarray-based screening,
circYWHAZ (
circ_0005962) was identified as one of the upregulated circRNAs in lung adenocarcinoma (LUAD)
[20]. The knockdown of
circYWHAZ by siRNAs significantly suppressed the proliferation of LUAD cells, implying that this circRNA can promote cell proliferation
[21]. Moreover,
circYWHAZ expression was also upregulated in plasma samples, which illustrates a good diagnostic value for LUAD patients (AUC = 0.73). After surgical resection,
circYWHAZ expression in the plasma decreased considerably, which suggests that
circ_0005962 is potentially a good noninvasive biomarker for LUAD diagnosis
[21]. miRNA-target prediction and functional enrichment analysis showed that
circYWHAZ might function as a miRNA sponge to regulate LUAD development, which needs further validation.
circACP6 (
circ_0013958) was also upregulated in LUAD tumors compared with nontumor tissues, which was validated by microarray and RT-PCR
[22]. High expression of
circACP6 was associated with the TNM stage (
p = 0.009, Cox analysis) and lymphatic metastasis (
p = 0.006) in LUAD patients. Moreover, the plasma expression levels of
circACP6 distinguished LUAD from the control (AUC = 0.794, 95% CI = 0.703–0.912). Additionally, knockdown of
circACP6 inhibited the proliferation, migration, and invasion of LUAD cells. Mechanistically,
circACP6 functioned as a sponge against miR-134, which promoted the upregulation of cyclin D1, a target of miR-134.
circCXCR4 (
circ_0056616) was identified and detected as a CXCR4-related circRNA in LUAD cells and exosomes
[23]. Plasma exosome levels of
circCXCR4 were lower in LUAD patients with TNM stage III–IV or with lymphatic metastasis than in those with stage I–II or without metastasis, respectively. This suggests that
circCXCR4 might suppress the progression and metastasis of LUAD. Indeed, plasma exosomal
circCXCR4 represents a good biomarker to diagnose lymphatic metastasis of LUAD (AUC = 0.812, 95% CI = 0.720–0.903), which also needs to be validated in a larger group of patients.
Through the exoRBase database (
http://www.exorbase.org; accessed on 3 June 2020),
circSATB2 (
circ_0008928) was selected as a highly expressed circRNA in cancer exosomes
[24]. The expression of
circSATB2 was higher in lung cancer cells than in normal bronchial epithelial cells. Furthermore, overexpression and knockdown experiments showed that
circSATB2 promoted the proliferation, migration, and invasion of lung cancer cells. Additionally, the packaging and transfer of
circSATB2 by exosomes influenced the proliferation and migration of the recipient cells.
circSATB2 directly bound to and inhibited miR-326, which in turn upregulated FSCN1, the presence of which has been reported as a poor prognostic marker for NSCLC patients
[25]. Therefore, upregulation of FSCN1 by
circSATB2 via sponging miR-326 represents a potential mechanism through which
circSATB2 promotes NSCLC progression. In addition, serum exosomal
circSATB2 expression was higher in NSCLC patients with metastasis than in those without, demonstrating a good diagnostic value for metastatic NSCLC (AUC = 0.797, 95% CI = 0.698–0.896).
In contrast, RNA-seq profiling demonstrated that
circ_0102537 was one of the downregulated exosomal circRNAs in LUAD, which was also retrieved from a microarray database (GSE101586). Moreover,
circ_0102537 was confirmed by quantitative RT-PCR to be downregulated in both plasma exosomes and tissues from LUAD patients.
circ_0102537 knockdown by siRNAs promoted the migration and invasion of lung cancer cells and enhanced the expression of EMT markers such as N-cadherin, Snail, and Vimentin. This suggests that
circ_0102537 might function as a tumor suppressor; however, the functional mechanism has not been presented
[26]. Although many circulating circRNAs have been linked to lung cancer progression so far, further validation with more diverse groups of patients and in-depth mechanistic studies should be performed.
3.2. Anticancer Drug Response
Over a long period, numerous studies have been conducted to find predictive markers for sensitivity to EGFR inhibitors
[27], and several circulating circRNAs have been proposed as candidate markers. Microarray analysis of plasma RNAs from NSCLC patients sensitive or resistant to gefitinib, an EGFR inhibitor, revealed that 1377 circRNAs were differentially expressed between the two groups
[28]. Among them,
circZNF91 (
circ_0109320) was upregulated in the gefitinib-sensitive group. The plasma levels of
circZNF91 could distinguish the gefitinib-sensitive group from the resistant group (AUC = 0.8054) and were associated with better progression-free survival in NSCLC patients treated with this EGFR inhibitor. Overall,
circZNF91 could be a predictive biomarker of the sensitivity to gefitinib treatment in NSCLC patients after comparative verification with other parameters in a wider and larger group of patients.
circC3 (
circ_0002130) increased in NSCLC cells that acquired resistance to the EGFR tyrosine kinase inhibitor, osimertinib
[29].
circC3 knockdown inhibited proliferation, glycolysis, and tumor growth in osimertinib-resistant lung cancer cells
[30].
circC3 acted as a sponge against miR-498 to upregulate its targets, GLUT1, HK2, and LDHA, which are glycolysis-related proteins. Furthermore, an increase in
circC3 was detected in serum exosomes from osimertinib-resistant NSCLC patients with respect to those from osimertinib-sensitive patients.
circC3 provided a good diagnostic value to predict the efficacy of osimertinib treatment in NSCLC patients (AUC = 0.792, 95% CI = 0.676–0.909), suggesting circulating
circC3 as a novel biomarker. A combination of two or more circulating circRNAs with other variables such as EGFR mutations and gene copy number
[27] would be a better biomarker for predicting the sensitivity to EGFR inhibitors.
circCNIH4 (
circ_0000190) and
circSHPRH were identified by RNA-seq to be upregulated in lung cancer cells compared with normal bronchial epithelial cells. They were also detected in conditioned media from lung cancer cells and in blood plasma samples by droplet digital PCR
[31]. Furthermore, their plasma levels exhibited a poor response to immunotherapy, which might be due to the upregulation of soluble PD-L1 caused by these circRNAs
[32]. Even though the detailed mechanism underlying the interplay between these circRNAs and antitumor immunity is still elusive, along with PD-L1 expression, their plasma levels could predict immunotherapy efficacy in lung cancer patients.
In-depth analysis of two GEO microarray datasets (GSE101684 and GSE101586) identified circRNAs highly expressed in LUAD samples compared with normal tissues
[33]. Among them,
circ_002178 promoted PD-L1 expression via sponging miR-34a.
circ_002178 was also highly detected in plasma exosomes from LUAD patients compared with those from healthy volunteers, and exosomal
circ_002178 had a significant diagnostic value for LUAD (AUC = 0.9967). Intriguingly,
circ_002178 was transferred from cancer cells to CD8
+ T cells via exosomes and then promoted PD-1 expression via sponging miR-28-5p. This indicates that
circ_002178 would be a good target for immunotherapy, since it can modulate the expression of PD-1/PD-L1 in LUAD. Circulating circRNAs are highly likely to be exploited as markers for predicting the responses to anticancer drugs once their mechanisms of action are confirmed and their efficacy is validated in more diverse patients.
3.3. Cancer Diagnosis and Prognosis
As noted previously, numerous circulating circRNAs (
circFARSA,
circYWHAZ,
circACP6,
circSATB2,
circZNF91,
circC3, and
circ_002178) have significant diagnostic value and are associated with prognosis in lung cancer patients. In addition, RNA-seq and subsequent RT-PCR validation confirmed that
circCD226 (
circ_0047921) and
circRALB (
circ_0056285) were downregulated, while
circATXN7 (
circ_0007761) was upregulated in serum exosomes from NSCLC patients
[34]. The combination of these three circRNAs provides a noteworthy diagnostic tool, which distinguishes NSCLC from healthy control (AUC = 0.919, 95% CI = 0.877–0.962) or other lung diseases, and their expression levels were associated with NSCLC progression.
Plasma
circCNIH4 demonstrated diagnostic potentials in lung cancer patients at all TNM stages (AUC = 0.95 for stage I–IV, AUC = 0.896 for stage I–II, and AUC = 0.96 for stage III–IV)
[31]. Patients with high plasma levels of
circCNIH4 exhibited poorer overall survival rates than those with low levels. Mechanistically,
circCNIH4 could modulate the EGFR/ERK pathway by sponging miR-142-5p
[35].
circPVT1 was also upregulated in tissues and sera from NSCLC patients. Both tissue and serum levels of
circPVT1 showed diagnostic potential, distinguishing NSCLC patients from controls (AUC = 0.803 and 0.794, respectively)
[36]. The knockdown of
circPVT1 by siRNAs suppressed proliferation, migration, and invasion and promoted apoptosis in lung cancer cells.
circPVT1 facilitated E2F2 signaling by functioning as a sponge against miR-125b. Even though researchers did not present the effect of
circPVT1 on the survival or prognosis of NSCLC patients recruited, they proved that
circPVT1 can be used as a diagnostic marker and elucidated its working mechanism in NSCLC.
The expression levels of several circulating circRNAs are associated with major mutations found in lung cancer. For example,
F-circEA, but not its host linear mRNA, could be detected in EML4–ALK
+ lung cancer plasma; thus, circulating
F-circEA would be a novel biomarker to detect EML4–ALK fusion and to determine an effective treatment for EML4-ALK
+ patients
[16]. LUAD patients with high plasma expression of
circBNC2 (
circ_0086414) were revealed to harbor EGFR mutations more frequently than those with low expression (
p = 0.001)
[21], suggesting that plasma
circBNC2 would be a companion diagnostic marker for EGFR tyrosine kinase inhibitors. Considering the examples described above and the stable structure of circRNAs, circulating circRNAs can be novel biomarkers for diagnosis, prognosis, and treatment monitoring in lung cancer patients.